How to forecast exchange rate, an unanswered puzzle

How to estimate the exchange rates is still an unanswered puzzle. Many economists have found various models to forecast the exchange rate, but these models still have some limitations. Some models achieve obvious successes for particular years, but perform very poor in another years. This paper compare three forecast models for different time horizon, and introduce a survey about the dealers´ opinions for estimating the exchange rates. The first model is macroeconomic balance approach. N, Dvornak, M. Kohler and G. Menzies (DKM) (2003) found this model to forecast the exchange rate over the medium-run. The next two models are estimating the exchange rate for the long run. First, R. Lyons applies dispersed information approach. Second, C. Neely and L. Sarno (NS) (2002) evaluated the different methods of using monetary fundamentals. The survey made by T. Hutcheson (2003) explained dealers´ opinion to estimate exchange rate. All the approaches I mentioned in this paper have some limitations. The movements of the exchange rates are very difficult to forecast, because the factors that influence the exchange rate always change over time and also there are many unpredicted events can affect on the movement of the exchange rates. In short, further researches still need to do to forecast the movement of the exchange rates.